The Agent Skills Directory

[SAFE]: The skill consists of purely instructional text guiding the AI agent through a code review workflow. No scripts, executable code, or network-enabled commands are included in the skill definition.
[COMMAND_EXECUTION]: The instructions refer to using a local CLI tool with flags such as --base-commit and --base. These are standard parameters for code analysis and version control tools used in a development environment.
[INDIRECT_PROMPT_INJECTION]: The skill involves analyzing code findings and repository output. While this represents a surface where untrusted data (code to be reviewed) enters the context, this is the primary and intended function of the skill. The instructions emphasize evaluating findings against repository rules and intent, which aligns with safe operating practices.

sociuu-coderabbit